FAQ

Frequently Asked Questions

Got questions? Get in touch.

What can Ragen actually do for my business?
Ragen turns your company documents into an AI-powered knowledge base that anyone on your team can query instantly.

Faster Decisions: Your team gets accurate answers in seconds instead of hours spent searching through files and asking colleagues.

Better Customer Experience: Deploy an AI chatbot that knows your products and policies – available 24/7 with personalized, context-aware responses.

Higher Productivity: Automate information retrieval and repetitive tasks so your team can focus on work that actually matters.

Read more: What is RAG and how does it work in a company? →
How quickly can we get started?
Fast. Book a 30-minute demo and we'll show you how it works with your specific use case. Most teams are up and running within days, not months.
Is Ragen safe for my company data?
Specifics rather than declarations. Once encryption is enabled, conversation content is encrypted with AES-256-GCM under a separate key per thread, and the master key stays in your own KMS. Personal data can be masked automatically. Access is controlled at the team and role level, and down to individual files and folders. The audit log records who changed what, with before-and-after state. In an on-premise installation with a locally hosted model, none of it leaves your network. Connect a commercial model with your own key and the queries themselves go to that provider – your choice, configured per organization.
How many people can use Ragen at the same time?
We do not impose a user limit – Ragen runs on your infrastructure, so scale depends on the resources you give it. In every installation you can create teams, assign roles, and control who has access to what information.
How is Ragen different from ChatGPT?
ChatGPT is a general-purpose AI assistant trained on the internet. Ragen is a company AI assistant that knows your business specifically.

A few things that make it meaningfully different:

It answers from your documents – not general internet knowledge. Upload your procedures, policies, contracts, and knowledge base. Ragen answers from those.

It finds the right document even when you phrase the question differently from how it was written. Ask "how do I cancel?" and Ragen finds the policy that says "terminating your plan." Ask for invoice FV-2024-001 by number and it finds exactly that document – not a conceptually similar one.

Long documents don't get lost. A 200-page contract is as findable as a one-page memo, because Ragen understands what each document is about as a whole – not just which words appear inside.

Plus: access control per department, integrations with your existing tools, and the ability to deploy a customer-facing chatbot – all from one platform.
Can I track how my team uses AI?
Yes. Ragen shows AI usage per assistant, per user and per model – calls, tokens and estimated cost over time – plus knowledge analytics: questions, unique users, answer ratings and the most cited documents. This helps optimize costs and measure ROI on your AI investment.
How is the MCP gateway different from MCP integrations?
Integrations are the sources Ragen pulls knowledge from – Google Drive, HubSpot, ClickUp and the rest. The gateway is traffic in the other direction: Ragen hands knowledge out to agents outside the platform, for instance those in Claude Desktop or Cursor.

Read more: Ragen for agents →
Do I have to use the Ragen AI Assistant to use the gateway?
You do not have to use its chat. An assistant in Ragen is the container for knowledge and permissions, but the ragen_search_knowledge_base tool returns document fragments only – your own agent composes the answer from there.
Where can I deploy Ragen?
On your own infrastructure. You install Ragen on your own servers or in your AWS, Azure or Google Cloud account, which gives you complete control over your data. Your team uses the AI Assistant, and your customers interact with the Chatbot on your website. If you would rather not do it yourself, we deploy it for you.
What AI models does Ragen support?
Any model a route in your configuration points at – including one running on your own GPU. The application calls providers directly: commercial models connect through your own keys (OpenAI, Anthropic, Google, Azure OpenAI, AWS Bedrock); open-source models run locally through Ollama or vLLM. The entire RAG pipeline can stay on-premise with nothing leaving your network, provided all four model roles (answering, rephrasing, summaries, embeddings) run on local models, Docling runs with DOCLING_STRICT=1 (without it, a PDF Docling cannot parse may be sent to an external model) and the content-moderation rule stays off. Any OpenAI-compatible endpoint, Portkey or LiteLLM included, plugs in as one more route. By default, semantic search uses an open-weight embedding model hosted in the EU; reranking is optional. Switching provider is a config change, not a migration – and which models are available to whom is set in the admin panel.
What tools does Ragen integrate with?
Ragen connects with Google Drive, HubSpot, ClickUp, Fireflies.ai, Google Calendar, and many more. Your AI assistant has access to up-to-date data from the tools you already use – no manual copy-pasting required.
Can I customize the assistant for my industry?
Yes. Ragen learns from your documents, procedures, and knowledge bases. You can create specialized assistants – e.g., one for HR, another for Sales – each with its own instructions and access to the right knowledge sources.
What if Ragen can't find the right document?
This is something we've specifically engineered against. Ragen searches your knowledge base from multiple angles simultaneously – by meaning, by exact keywords, and by overall document topic. So if your team asks "how do I cancel?" and your policy document says "terminating your plan," Ragen still finds it. If someone types an invoice number or a product code, Ragen finds the exact document – not a conceptually related one. And long documents (contracts, manuals, reports) are as findable as short ones, because Ragen generates a topic summary for each document at upload time. In practice: if the document is in your knowledge base, Ragen finds it.
Our documentation is in disarray. Does it make sense to start?
It does, just not by uploading everything at once. Ragen works with what you give it, so scattered information on the way in gives you faster access to the same scattered information. That's what Ragen Brain is for: it extracts knowledge pages from your documents, and someone at your company approves them before they reach the assistant.
Does an agent connected through the gateway bypass permissions?
No. Permissions work at document level and hold the same way for a person in a browser and for an agent over MCP. The agent works inside its organisation's API key and sees exactly the documents a person would see.
Can Ragen replace a local AI agent such as Hermes Agent?
They solve different problems and work well together. Hermes Agent is a personal agent for one person: it carries out tasks on a server, works in the terminal and on files, remembers its user and writes new skills for itself. Ragen is a knowledge platform for the whole company: it answers from your documents and cites its sources, enforces folder and file permissions, keeps an audit log and masks personal data.

If you want local AI for many employees, on your own server with your own models, choose Ragen – it runs on models served through Ollama or vLLM, just as Hermes does. If someone on the team wants a personal agent, they can run it alongside Ragen, and if the agent supports MCP, it can draw on company knowledge through the gateway, within the same permissions.

Read more: Ragen for agents →